While unfolded states of proteins were previously thought to be largely random, evidence of specific structure is accumulating. Even at low populations this type of ordering could have direct bearing on protein stability, folding mechanism, and rate, as well as the misfolding and aggregation that are implicated in many human disorders. In most cases, the weak nature of the residual structure and low populations of misfolded states under physiological conditions preclude direct and detailed characterization of these ensembles using experimental techniques. Simulations have begun to provide data complementary to experiments, since they can provide structural detail with single molecule resolution on a time scale inaccessible to most experiments. However, simulations fail in many important cases for multiple reasons. This proposal outlines continued development of simulation algorithms and force fields to enable successful application to the study of unfolded states, with direct validation against experimental data obtained through established collaborations. A strong emphasis is placed on conformational sampling; a major component of the project involves development of a novel sampling approach that provides improved convergence of ensemble data with explicit inclusion of solvent at a significantly reduced computational cost. Simulations will be performed on several model systems of varying complexity. In each case, specific mutations will probe interactions that are hypothesized to be involved in determining the structure or stability of the native fold. Other mutants probe the effect on stability of interactions that we have observed in the unfolded state. These studies will provide useful insight into these important model systems, and will also provide valuable and critical feedback on the performance of our force fields and sampling methods. The next phase of the research involves characterization of unfolded ensembles for each model system, and generation of experimentally testable hypotheses about the nature of any residual structure. For each of the model systems, recent experiments suggest that residual structure may exist in the unfolded state, and our simulations will provide important models for interpretation of this data. Additional studies of the unfolded state and the role of entropy in folding will be performedthrough replacement of flexible glycine and rigid proline with alanine, investigating the effect on unfolded state entropy and free energy of folding.